National surveys, NIH-funded studies, and other studies often rely on children to provide 24-hour dietary recalls (24hDRs), but validation studies show that children's recall accuracy is poor. A 24hDR obviously covers 24 hours of intake, but the target period (i.e., period for which intake is to be reported) may be the previous day (i.e., midnight to midnight of the day before the interview) or the prior 24 hours (i.e., from 24 hours before the beginning of the interview up to when the interview starts). The interview time (i.e., time the interview occurs on a day) maybe morning, afternoon, or evening. In general, as time elapses between an event and its recall, reporting accuracy declines; thus, the sooner an event is recalled, the more accurate the report is. Aside from a small pilot project, no validation study has evaluated the impact of manipulations of target period and interview time on the accuracy of children's 24hDRs. The hypothesis is that accuracy can be maximized by interviewing children as close in time as possible to the preponderance of their daily intake by asking them about the 24 hours immediately preceding the interview. The goal of the 4-year project is to investigate the relationship of reporting accuracy in children to manipulation of target period and interview time. Observations of school breakfast and school lunch will be used to validate these portions of the 24hDRs. The Primary Aim is to determine which of 6 combinations of 2 target periods and 3 interview times elicits the most accurate dietary reports from children. Each of 372 randomly selected 4th-grade males (M) and females (F) will be observed eating school breakfast and school lunch and interviewed using 1 of the 6 interview protocols, with 62 children (31 M, 31 F) per protocol. Reports will be compared to observations, and accuracy will be assessed at the item-, amount-, and nutrient-levels. Because an effect of observation on reporting performance would limit the ability to generalize the findings to unobserved children, the Secondary Aim is to ascertain whether observation of school meals influences dietary reporting. Each of 180 randomly selected 4th-grade children who are not observed eating the 2 school meals will be interviewed using 1 of the 6 interview protocols, with 30 children (15 M, 15 F) per protocol. Key information will be compared between reports from these 180 children and the 372 children observed and interviewed for the Primary Aim. Optimizing the conditions under which children report their dietary intake could meaningfully improve children's recall accuracy. Results from this methodological project could have major implications for how 24hDRs are obtained from children in future national surveys, NIH-funded studies, and other studies.